Support TF 2.16

[uri-granta]

Related issue(s)/PRs:

Summary

Add support for Tensorflow 2.16.1. Since this uses Keras 3 as the default keras, I've had to switch all uses of tf.keras (including hidden ones like tf.optimizer) to gpflow's tf_keras version-agnostic shortcut.

I've also put a maximum supported TF version, to prevent people trying to install trieste with unsupported versions in the future (since TF minor versions are offten not backwards compatible).

And I've removed support for Python 3.7 (which is already end-of-life) and 3.8 (which ends of life in less than a month) and switched the minimum supported version to 3.9.

Fully backwards compatible: yes

PR checklist

• The quality checks are all passing
• The bug case / new feature is covered by tests
• Any new features are well-documented (in docstrings or notebooks)

[khurram-ghani]

LGTM. (Had a quick chat offline).